Hey everybody, Solomon from team Docker. We're all lurking here and are happy to answer questions.
Just so you know I conzider Hacker News to be a key part in Docker's success. Your feedback, both positive and negative, made us feel like we weren't crazy and were building something that mattered. I hope in the end you will feel we live up to the expectations :)
We have lots of really cool things in store this year, I hope you will like them.
Happy hacking on behalf of the entire Docker team and contributors.
Hi Solomon. Congrats on the round. We're building on top of Docker a reproducable genomics platform at Seven Bridges Genomics. For us, Docker makes science and healthcare reproducible. Let's brainstorm on how we can bring wider visibility to Docker in that vertical :)
I'm working in the genomics space as well. We have an open source interface for filtering and examining variants for clinical diagnostics called Varify. We presented a poster at ASHG and are trying to get the open source code we have out on github documented and installable (https://github.com/cbmi/varify). My experience is that most of the analysis companies have the final variant display, filtration, and reporting as a gap in their capabilities and don't have the time or resources to devote to it 100%. Clinical diagnostics (where we are focused) has very different usage scenarios from discovery work, so it may not be a good fit depending on your focus.
Regardless, I'd love to hear about your experience with Docker and NGS pipeline work. Looked at your profile but there was no email. Shoot me a note if you want to chat (email is in my profile).
From what I understand, Docker looks like it can solve a lot of the practical issues involved with a Service Oriented Architecture (SOA).
Have you guys looked at targeting the enterprise market? Lots of enterprises that I'm talking to are looking into building their own private clouds for their SOA, and Docker could make that much easier.
From what I've seen on the technical side you just need rock-solid stability (or a strategy to achieve that via distributed architecture) and good management tools. Meat-space requires more work/investment (support, training, certification etc).
Congratulations. I'm glad to hear that HN feedback was a factor in making the team feel they were building something that mattered. I know there were always plenty of comments in threads about Docker that were dismissive, but count me among the people that were always pleased to see containerization on linux made trivial.
Given that Docker is licensed under Apache 2.0 which allows the venture backed Docker Corp. to create a closed proprietary version of the software... will outside contributors be compensated?
This is Ben, the CEO of Docker, and Solomon's partner in crime. Being based on Apache 2.0 (and adopting an open design process, having a large number of non-Docker Inc. maintainers, having over 300 contributors) should be read as a strong signal that we will not go proprietary or even "open core." Anyone can use and modify Docker under Apache 2.0 without fear of "copyleft" or similar restrictions, which creates a strong disincentive for us to abuse our position by going proprietary. Our business plans, as stated in the blog, are around commercial support and hosted services. (In addition, being Apache vs. GPL makes it much easier for enterprises to adopt us, and much easier to get integrated into projects like OpenStack.)
No :) If we released a closed-source version we would get forked in a instant. It would be an extremely stupid thing for us to do.
We've stated repeatedly that Docker inc. will make money by 1) offering support and professional services to enterprise customers, primarily through channel partners like Red Hat and others; and 2) offering hosted managed services for things like image hosting, orchestration, container tracking etc. Since we have been operating a PaaS at large scale for several years now, we know how to sell hosted services and it makes sense for us to focus on that.
"While Docker itself will always remain free and open source, we are busy gearing up for the launch of both hosted services around Docker and commercial support for Docker."
You have it backwards. If it was GPL, that would prevent other people from creating closed proprietary versions of their software. GPL or AGPL would prevent competition. Since they are the copyright owner they get unrestricted use to the software even if the rest of us had to follow the GPL.
Awesome work. Quick question, how closely do you guys plan on working with (or if at all) the CoreOS guys? I'm really seeing the future of all servers running ubiquitously on CoreOS + Docker setups. Thoughts?
No thanks. I would prefer that Docker work equally well on as many base operating systems as possible. Having any form of coupling would turn Docker into another locked in platform which is exactly the opposite of the way I see the future of deployments.
I mean seriously you really want a single vendor owning the base OS of all servers ?
I'm working on an itty bittier jeos vm that allows basically any linux userland via docker (rhel, deb, arch etc). It'll ship with zfs and grsecurity on a long-term stable kernel. Completely open source, hackable (core src packages similar to arch PKGBUILDs) and rebuildable from source.
During our alpha period, Chaos Monkey (i.e. random reboots) is built in and will give you plenty of opportunities to test out systemd.[1]
Great. While I understand why they do this, I don't want to have to setup a cluster with automatic failover just to try it out.
And then there is etcd. I'd love to use etcd - all I want is a minimal Docker image with etcd installed. When I looked, it seemed like I had to either build it myself, or use a CoreOS image.
I know it's alpha software, but that isn't a great experience.
Boot2Docker[2] OTOH: one 24 M image, burn it to a USB stick, boot it (in about 5 seconds!) and now you have the perfect Dockerhost.
To answer your question, our priority is to make sure docker works great on every distro, large or small, old or new. As some of the answers below will show you, sysadmin-world is quite a fragmented place, and I don't think we should expect "one distro to rule them all" situation. Instead I expect we will see more and more niche players solving a specific set of problems for a specific group of sysadmins, catering to certain tools preferences (openstack, mesos, hadoop/yarn, cloud foundry, juju, systemd, and so on). I think CoreOS is showing the way for a lot of those future distros. We want to make sure docker can be a good citizen in all of them.
My entire focus at Docker is building great relationships with the companies building software that use Docker as a part of their stack, whether it be for CI/CD purposes all the way down to building tools like PaaS.
So far we have been incredibly grateful for all of the interest and all of the amazing projects that are out there delivering real value with Docker today. The ecosystem continues to surprise and delight.
Please get in touch (nick@docker.com) if this is something you're interested in exploring. I'm also hiring if these are conversations you'd like to be having with others.
Do you guys have any plans to give BSD systems (FreeBSD has Jails perfectly suited for Docker) some love, given the amount of money you were able to get? :)
Yes :) We are working on pluggable execution drivers. The current driver (lxc) will be replaceable by eg. libvirt, plain chroot and eventually zones, jails etc.
There is actually someone experimenting with Jails support already (can't remember his irc handle atm). If you want to come by irc or open a github issue, we can put you in touch.
I've wanted to build something like Docker for FreeBSD Jails for a long time. I've been watching the Docker project since it was announced, and I spent this past week almost exclusively working with Docker and testing it for a possible production deployment (hope you get have a 'production ready' release soon :). In short, I would love to have Docker on FreeBSD!
Just so you know I conzider Hacker News to be a key part in Docker's success. Your feedback, both positive and negative, made us feel like we weren't crazy and were building something that mattered. I hope in the end you will feel we live up to the expectations :)
We have lots of really cool things in store this year, I hope you will like them.
Happy hacking on behalf of the entire Docker team and contributors.